iOn June 7, 2023, Singapore’s Minister for Communications and Information announced that AI Verify,[1] the world’s first-ever artificial intelligence (AI) governance testing framework and toolkit, will be made available to the open-source community.
AI Verify was launched as a minimum viable product for international pilot in 2022. It is aimed at helping organizations validate the performance of their AI systems through internationally recognized governance principles[2] and standardized tests.
The principles on which AI Verify is premised are as follows:
- Transparency
- Explainability
- Repeatability/reproducibility
- Safety
- Security
- Robustness
- Fairness
- Data governance
- Accountability
- Human agency and oversight
- Inclusive growth, and social and environmental wellbeing
AI Verify, a single integrated software toolkit operating within a user’s enterprise environment, allows users to conduct technical tests on their AI models and to record process checks. It also generates testing reports for these AI model tests, which facilitates transparency in AI development and deployment. Additional toolkits, to take into account sector-specific governance, for instance, can be built upon it.
Singapore has also launched the AI Verify Foundation,[3] a not-for-profit, multistakeholder body, for members to collaborate on AI governance and testing through information sharing and educational outreach of best practices.
Privacy World will continue to cover developments. For more information, contact your relationship partner at the firm.
[1] https://aiverifyfoundation.sg/
[2] For instance, those from the EU, the OECD, and Singapore.
